Regulations governing the operation of electric power systems in North America and many other areas of the world are undergoing major changes designed to promote competition. This process of change is often referred to as deregulation. Participants in deregulated electricity systems may find that their profits will greatly benefit from the implementation of successful bidding strategies. While the goal of the regulators may be to create rules which balance reliable power system operation with maximization of the total benefit to society, the goal of generation companies is to maximize their profit, i.e., return to their shareholders. The majority of the research described here is conducted from the point of view of generation companies (GENCOs) wishing to maximize their expected utility function, which is generally comprised of expected profit and risk. Strategies that help a GENCO to maximize its objective function must consider the impact of (and aid in making) operating decisions that may occur within a few seconds to multiple years;The work described here assumes an environment in which energy service companies (ESCOs) buy and GENCOs sell power via double auctions in regional commodity exchanges. Power is transported on wires owned by transmission companies (TRANSCOs) and distribution companies (DISTCOs). The proposed market framework allows participants to trade electrical energy contracts via the spot, futures, options, planning, and swap markets;An important method of studying these proposed markets and the behavior of participating agents is the field of experimental/computational economics. For much of the research reported here, the market simulator developed by Kumar and Sheble and similar simulators has been adapted to allow computerized agents to trade energy. Creating computerized agents that can react as rationally or irrationally as a human trader is a difficult problem for which we have turned to the field of Artificial Intelligence and Robotics; Some of our work uses GP-Automata, a technique which combines genetic programming and finite state machines, to represent adaptive agents. We use a genetic algorithm to evolve these adaptive agents (each with its own bidding strategy) for use in a double auction. The agent's strategies may be judged by the amount of profit they produce and are tested by computerized agents repeatedly buying and selling electricity in an auction simulator. In addition to the obvious profit-maximization strategies, one can also design strategies which exhibit other types of trading behaviors. The resulting strategies can be used directly in on-line trading, or as realistic models of competitors in a trading simulator;In addition to developing double auction bidding strategies, we investigate and discuss methods of an energy trader's risk. This can be done using such financial vehicles as futures and options contracts or through the inclusion of risk while judging strategies used in the market simulations described above. We discuss the role of fuzzy logic in the competitive electric marketplace, including how it can be applied in developing bidding strategies. Since competition promises to drive the power system closer to its operating limits, improvements in measurement and system control will be important. We provide an example of using fuzzy logic to do automatic generation control and discuss extensions that would make it superior to traditional controllers. Since the GENCO's forte is primarily generating electricity, we examine unit commitment and discuss how to update it for the competitive environment. We discuss the role of unit commitment in developing bidding strategies, as well as, the role of bidding strategies in solving the unit commitment problem. Depending on the market structure adopted by a particular location, large amounts of bidding data may be available to regulators or market participants. Ideally, regulators could use this data to verify dig the market is efficient. Market participants with access to this data might gain an advantage over their competitors if they could somehow determine their competitor's bidding strategy. We outline methods of automatically inferring other participants' trading rules based on historical data. Much of the work described here should aid in the design of effective operating procedures, trading strategies and profitable portfolios for energy producers.